I have remarked that IE roots attached to close enough meanings differed between them only by the alternation of R/L or the reverse.
According to M. POKORNIK (old reconstructions, but sensible):


*bal-bal- babbling, unclear speech

*bar-bar- babbling, unclear speech

*bhel- glittering white > bheleg to shine
*bher- shining, bright brown > bherəg, bhrēg to shine, bright, white
*bhlēu- < bhel- to bloat, to swell, to blow up, to flow
*bhrēu- > bhreu-d to swell, to sprout
*gel- to curl, to form into ball, round
*ger- to curl to wind, to turn
hel- (*ghel-?) >ğhle- to shine, to glare, to glow, gold, yellow, bright colour
*ğher- (-g-), *ğhre- to shine, to beam, to shimmer
*kal- hard, callous, blister
*kar- hard
*kel- to cut
*ker- to cut
? *kwel- to turn, wheel + ? neck
? *kwer- dish, pot (? meaning of roundness, or turned pots?)
*krā-(-u), *krə-, *krŭ- to heap up, to pile together, heap, roof
*klā- to lade, to place, to heap up, to lay down
*mel-, *melə- dark colour, dirty
*mer-, *mor- > *moru-, *moro- to blacken, dark colour, dirt spot
? *pel- to fold > pleǩ- to plait, to ply, to weave
? *prā- to bend
*wel-, *wlē- to turn, to wind, round, voluble
*wer- to turn, to bend, to wrap


So a repetition of the same phenomenon cannot be due to pure hazard. We could think that this alternation is old among IE languages ; in today dialects, it’s true, it ‘s a constant and so is not so surprising; it doesn’t need exchanges between remote IE dialects because it occurs on little scale too. I suppose it occurs also among other groups of languages.
What is your opinion? Could you post some similar examples in diverse languages?